Evaluation of Webcam-Based Eye Tracking for a Job Interview Training Platform: Preliminary Results

In job interviews, eye gaze towards the interviewer is an important non-verbal behavior that is considered a trait for hirability of a candidate. Several virtual job interview training platforms include eye trackers to measure eye gaze to provide feedback on performance. Though useful, these eye tracking devices are often pricey and not always accessible. In this article, we explore several camera-based eye tracking methods and implement a webcam-based eye tracking algorithm to determine its suitability for potential integration in virtual job interview simulation platforms. We further use the gaze predictions for interview relevant regions of interest detection. Our study with 12 participants, 7 with eyeglasses and 5 without, shows that during calibration, eyeglasses play no significant role in the differences in mean calibration error. Results from the ROI detection, however, show a limitation that it is important to maintain the same head position and distance during multiple tasks after calibration. Overall, webcam-based eye-tracking has potential, to be integrated in virtual job interview training environments.